test: add mzml contains peak udf (#202)

* test: add mzml contains peak udf
* feat: fix test path
* test: use all features
* feat: fixup fixture tests

.github/workflows/test.yml

@@ -65,4 +65,4 @@ jobs:
         awslocal s3api put-object --bucket test-bucket --key test.fasta --body ./exon/test-data/datasources/fasta/test.fasta
 
         cargo test
-        cargo test --package exon --test sqllogictests
+        cargo test --package exon --test sqllogictests --all-features

exon/src/datasources/vcf/scanner.rs

@@ -142,7 +142,7 @@ impl ExecutionPlan for VCFScan {
         if let Some(projections) = &self.base_config.projection {
             config = config.with_projection(projections.clone());
         }
-        tracing::debug!("VCF starting scan with config: {:#?}", config);
+        tracing::trace!("VCF starting scan with config: {:#?}", config);
 
         let opener = VCFOpener::new(Arc::new(config), self.file_compression_type);
         let stream = FileStream::new(&self.base_config, partition, opener, &self.metrics)?;

exon/src/datasources/vcf/table_provider.rs

@@ -479,20 +479,26 @@ mod tests {
         let path = crate::tests::test_fixture_table_url("chr17/")?;
 
         let ctx = SessionContext::new_exon();
-        let registration_result = ctx
-            .register_vcf_file("vcf_file", path.to_string().as_str())
-            .await;
-
-        assert!(registration_result.is_ok());
+        ctx.sql(
+            format!(
+                "CREATE EXTERNAL TABLE vcf_file STORED AS INDEXED_VCF COMPRESSION TYPE GZIP LOCATION '{}';",
+                path.to_string().as_str()
+            )
+            .as_str(),
+        )
+        .await?;
 
-        let sql = "SELECT chrom FROM vcf_file WHERE chrom = '17' AND pos BETWEEN 1000 AND 1000000 AND qual != 100;";
+        let sql =
+            "SELECT chrom FROM vcf_file WHERE vcf_region_filter('17:1000-1000000', chrom, pos) AND qual != 100;";
         let df = ctx.sql(sql).await?;
 
         let cnt_where_qual_neq_100 = df.count().await?;
         assert!(cnt_where_qual_neq_100 > 0);
 
         let cnt_total = ctx
-            .sql("SELECT chrom FROM vcf_file WHERE chrom = '17' AND pos BETWEEN 1000 AND 1000000;")
+            .sql(
+                "SELECT chrom FROM vcf_file WHERE vcf_region_filter('17:1000-1000000', chrom, pos)",
+            )
             .await?
             .count()
             .await?;

exon/src/udfs/massspec/contains_peak.rs

@@ -43,9 +43,9 @@ use datafusion::{
 /// * `result` - A BooleanArray. The BooleanArray contains true if the spectrum
 ///    contains a peak within the tolerance and false otherwise.
 fn contains_peak(args: &[ArrayRef]) -> DataFusionResult<ArrayRef> {
-    if args.len() < 3 {
+    if args.len() != 3 {
         return Err(datafusion::error::DataFusionError::Execution(
-            "contains_peak takes at least two arguments".to_string(),
+            "contains_peak takes three arguments".to_string(),
         ));
     }
 

exon/tests/sqllogictests/slt/mzml-functions.slt

@@ -0,0 +1,26 @@
+control substitution on
+
+statement ok
+CREATE EXTERNAL TABLE mzml_table STORED AS MZML LOCATION '$CARGO_MANIFEST_DIR/test-data/datasources/mzml-pyoteomics/pyoteomics.mzML'
+----
+
+query I
+SELECT contains_peak(mz.mz, 200.0, 1.0) AS has_peak FROM mzml_table LIMIT 1;
+----
+true
+
+query I
+SELECT contains_peak(mz.mz, 0.0, 1.0) AS has_peak FROM mzml_table LIMIT 1;
+----
+false
+
+statement error
+SELECT contains_peak(mz.mz) AS has_peak FROM mzml_table LIMIT 1;
+
+query I
+SELECT bin_vectors(mz.mz, intensity.intensity, 200.0, 10, 1.0) AS bins FROM mzml_table LIMIT 1;
+----
+0,0,0,0,203667.40002441406,0,0,0,0,0
+
+statement ok
+DROP TABLE mzml_table